Summary
The release introduces TrajectoryJudge, which reviews a live run on a cadence and can steer it mid-run. It also adds filesystem traversal events, explicit OpenTelemetry decisions and a hosted engine definition for running repository-defined Pydantic agents in GitHub Agentic Workflows.
Why it matters
Live evaluation and intervention provide a practical control loop for agents whose failures cannot be detected reliably after completion.
Related coverage
- Fabrix.ai launches governed VibeOps for agentic SRE and operations
- Qwen Code 0.23.2 adds resumable multi-agent controls and safer remote sessions
- OpenAI Agents Python 0.22.2 adds image-tool support and closes a sandbox race